harness-evolver 2.5.0 → 2.5.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/package.json +1 -1
- package/skills/init/SKILL.md +44 -24
package/package.json
CHANGED
package/skills/init/SKILL.md
CHANGED
|
@@ -48,30 +48,50 @@ If NO eval exists:
|
|
|
48
48
|
|
|
49
49
|
**Tasks** (`tasks/`): If test tasks exist, use them.
|
|
50
50
|
|
|
51
|
-
If NO tasks exist:
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
51
|
+
If NO tasks exist, generate them. First, identify all relevant source files:
|
|
52
|
+
|
|
53
|
+
```bash
|
|
54
|
+
find . -name "*.py" -not -path "./.venv/*" -not -path "./.harness-evolver/*" | head -10
|
|
55
|
+
find . -name "*.json" -o -name "*.md" -o -name "*.txt" -o -name "*.yaml" -o -name "*.yml" | grep -v .venv | grep -v .harness-evolver | head -10
|
|
56
|
+
```
|
|
57
|
+
|
|
58
|
+
Then spawn testgen subagent with CONCRETE file paths (not placeholders):
|
|
59
|
+
|
|
60
|
+
```
|
|
61
|
+
Agent(
|
|
62
|
+
subagent_type: "harness-evolver-testgen",
|
|
63
|
+
description: "TestGen: generate 30 test cases",
|
|
64
|
+
prompt: |
|
|
65
|
+
<objective>
|
|
66
|
+
Generate 30 diverse test cases for this project. Write them to the tasks/ directory
|
|
67
|
+
in the current working directory.
|
|
68
|
+
</objective>
|
|
69
|
+
|
|
70
|
+
<project_context>
|
|
71
|
+
This project is at: {absolute path to project root}
|
|
72
|
+
Entry point: {the harness/agent file you identified, e.g., crew.py or pipeline/moderator.py}
|
|
73
|
+
Framework: {what you detected — CrewAI, LangGraph, etc.}
|
|
74
|
+
</project_context>
|
|
75
|
+
|
|
76
|
+
<files_to_read>
|
|
77
|
+
{LIST EVERY .py file and data file you found above — use ABSOLUTE PATHS}
|
|
78
|
+
Example:
|
|
79
|
+
- /home/rp/Desktop/test-crewai/crew.py
|
|
80
|
+
- /home/rp/Desktop/test-crewai/README.md
|
|
81
|
+
</files_to_read>
|
|
82
|
+
|
|
83
|
+
<output>
|
|
84
|
+
Create directory tasks/ (at project root) with 30 files: task_001.json through task_030.json.
|
|
85
|
+
Format: {"id": "task_001", "input": "...", "metadata": {"difficulty": "easy|medium|hard", "type": "standard|edge|cross_domain|adversarial"}}
|
|
86
|
+
No "expected" field needed — the judge subagent will score outputs.
|
|
87
|
+
Distribution: 40% standard, 20% edge, 20% cross-domain, 20% adversarial.
|
|
88
|
+
</output>
|
|
89
|
+
)
|
|
90
|
+
```
|
|
91
|
+
|
|
92
|
+
Wait for `## TESTGEN COMPLETE`. If the subagent fails or returns with no tasks, generate them yourself inline (fallback).
|
|
93
|
+
|
|
94
|
+
Print: "Generated {N} test cases from code analysis."
|
|
75
95
|
|
|
76
96
|
## Phase 3: Run Init
|
|
77
97
|
|